Artificial Intelligence-Based Video Saliency Prediction: Challenges and Trends

Video saliency prediction (VSP) aims to identify regions in videos that attract human attention and gaze. In the past, researchers have conducted extensive studies on VSP, establishing various video saliency datasets and prediction models. Leveraging the powerful end-to-end learning capabilities of deep learning techniques and the availability of large-scale video saliency datasets, the performance of saliency prediction models has significantly improved. Today, with the development of multimedia technologies, the task of VSP has generated numbers of promising directions, such as high dynamic range VSP and audio VSP, among others. This article focuses on the challenges of VSP in the context of multimedia technologies; reviews the research on video saliency, including video saliency datasets and prediction models; and then introduces potential research directions in conjunction with contemporary multimedia technologies.

Authors

• Jiongzhi Lin
• Baitao Huang
• Fei Zhou College of Electronics and Information Engineering, Shenzhen University, China Peng Cheng National Laboratory, Shenzhen, China Guangdong Key Laboratory of Intelligent Information Processing, Shenzhen, China Shenzhen Institute for Artificial Intelligence and Robotics for Society, China Guangdong-Hong Kong Joint Laboratory for Big Data Imaging and Communication, Shenzhen, China

Keywords

• Deep learning
• Adaptation models
• Reviews
• Streaming media
• Predictive models
• Market research
• High dynamic range
• Intelligent systems
• Convergence
• Context modeling
• Saliency Prediction
• Video Saliency
• Convolutional Network
• Convolutional Neural Network
• Dynamic Range
• Spatial Information
• Computer Vision
• Short-term Memory
• Visual Information
• Long Short-term Memory
• Attention Mechanism
• Large-scale Datasets
• Temporal Information
• Visual Attention
• Promising Direction
• Real-time Information
• Artificial Intelligence Applications
• Human Visual System
• Human Attention
• LSTM-based Model
• Optical Flow
• Attention Patterns
• Technological Advances
• High-resolution Video
• Video Content
• Audio Content
• Deep Learning Technology